Ben Ward

Tumblr 89789175

.

These issues exist because the ultimate nature of human taste is irrational and depends on factors impossible to capture with computer systems.   Interpersonal relationships, social pressures, locales, context in which music is encountered, meaning that individuals assign to a music encounter (has a girl you liked ever introduced you to music you just happened to also like?)

I’ve just purchased ‘“Volunteered” Civility & Professionalism’ by Soccer Team. It’s downloading direct onto my iPhone as I write. I purchased this album because I’ve been listening to ‘Recommendations’ Radio all afternoon on Last.FM, and heard another of Soccer Team’s songs.

The above quote is from Anthony Volodkin in Notes from the “Help! My iPod Thinks I’m Emo!” panel. It was interesting in his analysis of Last.FM’s social, but algorithmic, recommendation system, and how it can leave you stuck in the most popular music, and less likely to break far out of your comfort zone. By contrast, a more editorial recommender, derived from a trusted, human source will provide unpredictable variation; deliberate obscurities. A recommender based entirely on fellow consuming crowds can drown out individual expertise.

Anthony’s panel at SXSW was fabulous. Great contrasts between social and mechanical recommendation mechanisms, well presented and set you thinking.

Both methods are good, both methods work and have different ideal scenarios. I’m unconvinced that a single service can provide a magic hybrid of behavioural and editorial recommendation, and in a way I hope that it can’t be done. There is pleasure to be had from varying the way you consume music. The experience of listening to the same piece of music on iTunes verses vinyl extends deeper than the sound itself, but affects a great deal of what you do and how you behave whilst you listen. I treasure that variety. Choosing to consume recommended music from Last.FM or from Hype Machine is far less stark difference, but different nonetheless. Interfaces matter, variety matters. Why might I choose to play ‘Hysteria’ by Muse in Rock Band rather than just picking out the song in iTunes? Because I gain more from the variety of the experience than I lose by ruining the guitar solo on ‘expert’ difficulty.

Now, my opening example is a success story for Last.FM’s existing recommendation system. But there’s a twist. I’ve not been listening to my recommendation radio. I’ve been listening to the recommendations generated by Last.FM for my friend Leslie. The input data isn’t about me, and the output data isn’t ‘for’ me. But, I’ve added my own element of editorial control; I’ve made an active (if crude) intervention to the recommendation system, based on my feelings for another individual’s taste in music, rather than anything the computer could have inferred.

Right now user-driven augmentation of recommendation engines is a good short term improvement. Yes, this is a crude example, but anything that has you more actively engaged with what you’re listening to should be embraced. More refined interactions will come with time, I’m sure.

(Also, in answer to Anthony’s question quoted above: Yes.) Via: fascinated.fm.

You can file issues or provide corrections: View Source on Github. Contributor credits.